Case Study: 10,279 Merchant Alerts Cleared in 7 Weeks
Seven weeks instead of clients estimated 12 weeks. 10,000+ transaction-monitoring alerts. 675 merchants, one of which carried 950+ alerts on its own.
That was the backlog a regulated payments provider asked CAML to help clear on its card-acquiring portfolio. The work later went through the client's audit, and we were told the cases our team investigated were approved.
Here is how the project ran, what the numbers looked like, and what we would tell any compliance team facing the same queue.
The project at a glance
10,279 alerts reviewed
675 merchant investigations completed
1 to 972 alerts per merchant
About 1,470 alerts processed per week on average
Weekly capacity grown from 150 to almost 300 analyst hours
Average handling time per merchant down from 3.7 hours in week 1 to 1.7 hours in week 6
46 of 1,618 recorded hours spent on training
The brief
The client first contacted us in March through the form on our website, went quiet, and came back in July with a deadline. We started in mid-August.
For every merchant, an analyst reviewed the alerts raised over the lookback period, identified which rules had fired and why, assessed the activity against the merchant's business, ran screening, documented the investigation and wrote a conclusion. Cases that needed more went to escalation or a request for information (RFI).
The client kept the final decision on every case. CAML provided the investigation capacity and the documented findings, working on the client's own laptops, accounts and systems.
Why alert count matters more than case count
The average merchant carried about 15 alerts, but the average hides a lot. Many cases had one or two alerts and closed quickly. Merchants with more than 100 alerts were almost always at the higher-risk end of the portfolio, often obliged entities themselves, and they produced most of the escalations.
When a cluster of those arrived midway through the project, a per-merchant KPI stopped being fair. We reviewed it with the client and agreed a revised measure. Sticking to the old target would have pushed analysts to rush exactly the cases that deserved the most care.
How the team got up to speed
Before our analysts touched a case, we shadowed one of the client's investigators and recorded a complete review, from the first alert to the final conclusion. That recording became the team's main training asset. An SOP lists the steps. A recording shows the judgement calls, and where the SOP is silent, that matters more.
The learning curve lasted about two weeks. Analysts focused on quality and asked a lot of questions. From week three the client's KPIs were met, and the strongest analysts went well past them. By the end, the team was consistently delivering above what had been agreed.
Three changes helped along the way:
Internal KPIs per analyst on top of the client's general targets, so we could see quickly who was ready for more volume.
A move from several part-time analysts to a full-time core team once we knew who performed best.
An internal QA step, so every investigation had a second CAML reviewer before handover.
What 10,000+ alerts turned out to be
The large majority of investigations concluded as false positives. A relatively small share needed escalation, an RFI or reporting, and our analysts raised several internal suspicious activity reports to the client's compliance team for decision.
That ratio is normal for a retrospective merchant review. It is also why the write-up is the real product. Closing a false positive well means explaining why the activity makes sense, in a way an auditor can follow months later.
Weekly reporting calls and in-person meetings at the client's office kept both teams aligned on volume, handling time and open questions. Feedback went back to the analysts the same week.
Frequently asked questions
What does an outsourced transaction-monitoring team actually do?
It investigates alerts within your existing methodology: reviewing alert history, understanding which rules fired, assessing activity against the customer or merchant profile, screening, documenting and concluding. You keep the risk appetite and the final decisions.
How long before the team is productive?
In this project, about two weeks to reach the client's KPIs, with 46 of 1,618 recorded hours spent on training. We usually treat the first week as a training period billed hourly, which also sets the KPI baseline for the rest of the engagement.
What happens when the case mix changes mid-project?
We review the KPI with you. A merchant with more than 900 alerts is not comparable to one with a single alert, and the target should reflect that.
How do you control quality?
Every completed investigation gets a second review by a CAML reviewer before handover. Supervisors and Team Leads are ACAMS-certified, and analysts assigned to outsourcing engagements have more than five years of AML experience.
Do your analysts work in our systems?
Yes. In this project the team worked on the client's laptops, accounts and system access. Allow a few days for access to be set up.
How is a backlog project priced?
Three options: a fixed hourly rate on actual reported hours, payment per completed case against an agreed definition of done, or payment per processed item based on time allocations set during the training week.
Can I speak to this client?
We don't publish client names. Once a mutual NDA is in place, we can share the client's identity and a full reference.
Sitting on a transaction-monitoring backlog?
We can consult you, help you get automated, give you free advice and or place a team that we perform well. Here is a similar project we did recently, read 450 corporate onboardings in under three months.
Comments